Learning interacting dynamical systems with latent Gaussian process ODEs
Authors: Çağatay Yıldız, Melih Kandemir, Barbara Rakitsch
NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We empirically demonstrate that our model improves the reliability of long-term predictions over neural network based alternatives and it successfully handles missing dynamic or static information. Furthermore, we observe that only our model can successfully encapsulate independent dynamics and interaction information in distinct functions and show the benefit from this disentanglement in extrapolation scenarios. We exhaustively test our method on a wide range of scenarios varying in function complexity, signal-to-noise ratio, and system observability. |
| Researcher Affiliation | Collaboration | Ça gatay Yıldız University of Tübingen cagatay.yildiz@uni-tuebingen.de; Melih Kandemir University of Southern Denmark kandemir@imada.sdu.dk; Barbara Rakitsch Bosch Center for Artificial Intelligence barbara.rakitsch@de.bosch.com |
| Pseudocode | No | The paper does not contain any explicitly labeled pseudocode or algorithm blocks. |
| Open Source Code | Yes | Our Py Torch [43] implementation can be found in https://github.com/boschresearch/i GPODE (GNU AGPL v.3.0 license). |
| Open Datasets | Yes | Datasets We illustrate the performance of our model on two benchmark datasets: bouncing balls [44] and charges [2]. |
| Dataset Splits | No | The paper mentions training and test sequences, but does not provide explicit numerical details on dataset splits (e.g., percentages or counts for train/validation/test sets). |
| Hardware Specification | No | The paper only mentions 'GPU clusters' in the acknowledgements without specific hardware details (e.g., GPU model, CPU model, memory, or specific cloud instances). |
| Software Dependencies | No | The paper mentions 'Py Torch [43]' and 'ACA library [42]' but does not provide specific version numbers for these or other software dependencies. |
| Experiment Setup | No | The paper states 'Due to the space limit, we refer to the Supplementary Material for more detailed information about the experimental setup and comparison methods', and the main text does not provide specific hyperparameter values like learning rate, batch size, or number of epochs. |